How the ostrich got his thick knees

The evolution of multicellular organisms is based on changes in genes that control the behaviour of cells in the developing embryo as well as in the adult. Thus the relation between development and evolution is fundamental, and related studies come under the title of evolutionary developmental biology or, more familiarly, "evo-devo". The problems in this field cover not only the mechanism of embryonic development but how embryonic development itself evolved, the evolution of different life histories that include larval stages, and the relation between development and ecology. It is an exciting field and it would be helpful to have a good book covering all such issues but, alas, this book is not it.

Rather than being a coherent text it consists of a series of 50 brief articles in alphabetical order dealing with what the title claims are the key concepts in the field. There is virtually no relationship between the articles, and the book is clearly not meant to be read but rather to be used as a reference by someone who wishes to know a little about topics that range from animal phyla through articles on innovation, phylogeny and segmentation, to end with variation.

So how helpful are the individual articles? They vary a great deal in quality and many important issues are simply neglected. Nevertheless, the book contains useful information.

At least one article tries to cover the field as a whole - "Development, evolution, and evolutionary developmental biology" by Scott Gilbert and Richard Burian. As the authors point out, a key issue is how existing developmental mechanisms are modified in evolution to give innovations in form and function. It is unlikely that any new mechanism evolves at once, rather that there is, to paraphrase Franois Jacob, tinkering with what is already there to make novelty. Gilbert and Burian also discuss modularity, which is not an obvious concept but relates to structures in the embryo that can develop independently but have no anatomical counterpart in the adult - for example, somites that give rise to vertebrae and muscles and dermis. Even limbs can be thought of as modules.

A central issue is that of evolvability itself, considered by John Gerhart and Marc Kirschner. They suggest that this capacity to generate non-lethal selectable phenotypic variation from random mutation may have itself evolved. They consider a toolkit analogy with emphasis on modules in which there is weak linkage between the modular processes to each other.

Jonathan Slack discusses the the somewhat controversial issue of the phylotypic stage, at which, in vertebrates, the embryos are all rather similar. He also emphasises how similar vertebrates are to flies with respect to the key genes that specify the body plan. Another useful article is on epigenetics, which is a modification of gene activity similar to a mutation but not based on alterations in the DNA, rather on its chemical modification. This is important for imprinting, which turns off a specific gene in a maternal or paternal genome and is partly responsible for why reproductive cloning of humans is so filled with likely abnormal outcomes.

A somewhat neglected topic discussed here is the Baldwin effect, which Conrad Waddington, one of the founders of evo-devo, made much use of. The basic idea is that an environment-induced change can become genetically assimilated. This may seem like Lamarckism but it is not.

Waddington's examples relate to the thickenings of the skin on ostriches' knees. How could this have evolved? The Baldwin effect suggests that repeated kneeling on hard ground and associated mutations led to thickening of the knee skin where there was pressure on the skin. Later, even when there was no external signal, this thickening - its cellular basis having been established - became genetically determined, so that birds were born with thick-skinned knees.

Another of Waddington's ideas was canalisation, the evolution of buffering against the effects of both genetic and environmental perturbations of a developmental process.

Many animals have complex life histories. Consider the very important model for development, the fruit fly Drosophila , which carries its adult parts in its larva; they develop when the fly emerges from the pupa. How did this evolve? The book does not discuss it and even suggests that direct development (from egg to adult without a larval stage) is a specialised reproductive mode. The opposite is much more likely, and it can be argued that all larvae are intercalations into the development of a direct developing animal.

Of course, genes control development, and we are coming to terms with the relatively small number of genes in humans compared with other animals, such as worms and flies.

The key to complexity lies in where and when these genes are expressed - determined by the genes' control regions so that a given gene can be expressed in many different places and times in development. This is neglected in the book, which also gives insufficient attention to how gene activity gives rise to different forms: both pattern formation - based for example on positional information, and the forces involved in morphogenesis, which mould the shape of the developing embryo.

Probably the most difficult problem in evo-devo is the adaptiveness, and hence selective advantages, of the intermediate forms that lead to current animals. Think of the forms that may have given rise to limbs in our fishy ancestors - what advantage for them would have been a little bump on the flank? What about wings before a bird can fly? Not to speak of gastrulation itself, which is at the base of all animal development. How were the intermediate forms an advantage? You will not find the answers, or even the questions, in Keywords and Concepts in Evolutionary Developmental Biology .

Lewis Wolpert is professor of biology as applied to medicine, University College London.